Categorization of Santa Ana Winds with Respect to Large Fire Potential

Categorization of Santa Ana Winds With Respect To Large Fire Potential

Tom Rolinski1, Brian D’Agostino2, and Steve Vanderburg2

1US Forest Service, Riverside, California. 2San Diego Gas and Electric, San Diego, California.

ABSTRACT

Santa Ana winds, common to southern California during the fall through early spring, are a type of katabatic wind that originates from a direction generally ranging from 360°/0° to 100° and is usually accompanied by very low humidity. Since fuel conditions tend to be driest from late September through the middle of November, Santa Ana winds occurring during this time have the greatest potential to produce large, devastating fires when an ignition occurs. Such catastrophic fires occurred in 1993, 2003, 2007, and 2008. Because of the destructive nature of such fires, there has been a growing desire to categorize Santa Ana wind events in much the same way that tropical cyclones and tornadoes have been categorized. The Offshore Flow Severity Index (OFSI), previously developed by Predictive Services, is an attempt to categorize such events with respect to large fire potential, specifically the potential for new ignitions to reach or exceed 100 ha based on breakpoints of surface wind speed and humidity. More recently, Predictive Services has collaborated with meteorologists from the San Diego Gas and Electric utility to develop a new methodology that addresses flaws inherent in the initial index. Specific methods for improving spatial coverage and the effects of fuel moisture have been employed. High resolution reanalysis data from the Weather Research and Forecasting (WRF) model generated by the Department of Atmospheric and Oceanic Sciences at UCLA is being used to redefine the OFSI. In addition to the new methodology, social scientists from the Desert Research Institute have been contracted to evaluate how this index might best be conveyed to the user so as to maximize its effectiveness. This paper will outline the methodology for developing the improved index as well as discuss how it might benefit fire agencies, private industry, broadcast media groups and the general public.

1. Introduction or not a Santa Ana wind event occurred. This is a necessary process as it helps distinguish a true Santa From the fall through early spring, offshore winds, or Ana from the normal nocturnal offshore winds that what are commonly referred to as “Santa Ana” occur throughout the coastal and valley areas. winds, occur over southern California from the coastal mountains westward, from Ventura County During 21 through 23 October 2007, Santa Ana southward to the Mexican border. These synoptically winds generated multiple large catastrophic fires driven wind events vary in frequency, intensity, and across southern California (Moritz et al.2010). Most spatial coverage from month to month and from year notable was the Witch Creek fire in San Diego to year, thus making them difficult to categorize. County, where wind gusts of 26 m/s were observed at Most of these wind events are associated with mild to the Julian weather station along with relative warm ambient surface temperatures >= 18° C and humidity values of ≈ 5%. However, high resolution low surface relative humidity <= 20%. However, model simulations at 667 meters showed that wind during the late fall and winter months, these events velocities were much higher in unsampled areas tend to be associated with lower surface temperatures (Fovell 2012). This event generated an interest in due to the air mass over the Great Basin originating categorizing Santa Ana winds so that, with such an from higher latitudes. There are a variety of ways to index available, fire agencies and first responders, define a Santa Ana event through the analysis of local private industry, and the general public could be and synoptic scale surface pressure and thermal more informed about the degree of severity an event distributions across southern California (Raphael would have on the fire environment. This index could 2003). For our purposes, Mean Sea Level Pressure also help augment Fire Weather Watches and Red (MSLP) map types and surface wind speed Flag Warnings from the National Weather Service by observations will be the determining factors whether

providing value added information about an impending event.

The Predictive Services Unit, functioning out of the Geographic Area Coordination Center in Riverside, California, is comprised of several meteorologists employed by the USDA Forest Service. In 2009, Predictive Services developed the Offshore Flow Severity Index (OFSI), which categorizes Santa Ana Figure 1 ‐ A 7‐day forecast of the Offshore Flow Severity wind events according to the potential for a large fire Index (OFSI) displaying categories of Santa Ana winds. to occur (Rolinski et al. 2011). This unique approach areas which have been divided into three zones (Fig. addresses the main impact Santa Ana winds could 1). Zone 1 covers the southern portion of Ventura have on the population of southern California beyond and Los Angeles Counties. Zone 2 consists of Orange experiencing the casual effects of windy, dry County, as well as western Riverside and western San weather. Bernardino Counties. Zone 3 represents most of San Diego County (Fig. 2). These zones were chosen in San Diego Gas and Electric (SDG&E) is a regulated part based on the different offshore flow utility provider across the southern portions of characteristics that occur across the region. For Orange County and all of San Diego County. During instance, Santa Ana winds across Zone 1 and Zone 2 the October 2007 event, it was determined that power are primarily a result of offshore surface pressure lines were the cause of several large fires, including gradients (locally and/or synoptically) interacting the Witch Creek fire in San Diego County. The cost with the local terrain to produce gap winds through of these fires to date currently exceeds 2 billion Soledad Canyon, the Cajon Pass, and the Banning dollars. As a result, SDG&E is investing in fire Pass (Hughes and Hall 2010; Fovell 2012). These weather related research and technology to develop winds also tend to precede the Santa Ana winds that an enhanced warning system for dangerous wildfire occur across San Diego County by 12 to 24 hours. conditions. Having advance notice of the severity Across Zone 3, offshore winds take on a more and timing of a Santa Ana wind event would permit “downslope windstorm” characteristic driven largely SDG&E to prepare, monitor, and deploy its resources by the tropospheric stability (Fovell 2012). In for maximum effectiveness. In order to accomplish addition, these zone boundaries were developed this, SDG&E is partnering with the local partially around political boundaries, as well as meteorological and fire community on this project. around the news media broadcast markets that cover the area. The OFSI has been proven to be successful in terms of capturing the overall nature of a Santa Ana wind event. However its basic method in addressing complex issues such as time, topography, and fuel conditions was perceived to be overly simplistic by the scientific community and also by other users of the index. These issues have since been addressed and will be discussed in detail in the following section.

2. Background

A seven day forecast of the OFSI is currently being produced by Predictive Services on a daily basis for Figure 2 ‐ Map depicting OFSI Zones over southern the southern California coastal, valley, and mountain California. Favored wind corridors are indicated by yellow arrows.

Wind Velocity (mph) Wind Velocity (mph) 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 0 5 5

(%) 10 (%) 10 15 15 RH 20 RH 20 25 25

All Fires Fires >= 100 ha

Figure 4 ‐ Scatter plot displaying days in which a fire Figure 3 ‐ Scatter plot displaying days in which a large (any size) occurred. fire (100 ha) occurred.

3. A New Methodology A large fire is defined as the 95th percentile of daily largest fires occurring over the past 20 years across a) Large Fire Potential – Meteorological the region depicted in figure 2. Thus a large fire for Conditions Zones 1, 2, and 3 is 100 ha, which is typically when The potential for an ignition to reach or exceed 100 additional resources will be required from outside the ha depends on a number of components: e.g. various point of origin to suppress the fire. This large fire meteorological and fuel conditions, suppression definition was used to develop the OFSI, which strategy, topography, accessibility, and resource correlates historical fire information with surface availability. Current methods to evaluate fire wind speed and humidity from Remote Automated potential include various indices from the National Weather Stations (RAWS), to initially form a four Fire Danger Rating System (NFDRS) and from the tier categorical index (Rolinski et al. 2011). Scatter Canadian Forest Fire Danger Rating System plots of wind velocity vs. relative humidity show the (CFFDRS) (Preisler et al. 2008). The Fosberg Fire frequency distribution of ignitions and 100 ha fires Weather Index (FFWI) is one such index which is a respectively over the dataset period during Santa Ana function of wind speed, humidity, and temperature wind events (Figures 3 and 4). Comparing these plots with output values ranging from 0 to 100 (Fosberg with each other revealed natural breakpoints in the 1978). While the FFWI may show elevated output data that were used to define the OFSI categories. values for a Santa Ana wind event, it can also show This process was repeated for multiple RAWS to elevated values for any day therefore making it too determine a site that would be most effective in generic for our purposes. The initial concept of OFSI representing each zone. Stations near the coast was a first attempt to create an index more specific to sometimes indicated offshore events as being weaker Santa Ana wind events, but further studies resulted in and of shorter duration than the mountains. We found a new approach, which we term Large Fire Potential inland valley stations usually worked best in (LFP). Assuming an aggressive suppression strategy capturing the overall nature of the event. In the final is employed with adequate resource availability in an analysis, the Saugus, Corona, and Julian RAWS were easily accessible area where topography is uniform, used to represent Zones 1, 2, and 3 respectively (Fig. LFP becomes a function of the fuel and weather 2). However, due to the necessarily simplistic nature conditions preceding, during, and following the time of this approach, the spatial complexities inherent to of ignition. Supposing for the moment that fuels are offshore flow were not captured. In addition, the fully receptive to ignitions and will support large fire

OFSI employed a rudimentary scheme to address fuel growth, the weather component of LFP (e.g., LFP) dryness which did not take into account other critical during a Santa Ana wind event can be expressed by components of fuel moisture. Both of these aspects the following equation: revealed a weakness in the initial conceptual model, but significant improvements have been made since [1] then which will be discussed in the next section.

components: 1) dead fuel moisture, 2) live fuel where W is the 6 m wind velocity (mph) based on moisture, and 3) the state of green-up of the annual the anemometer height used by RAWS and other grasses. Each of these components is complex and stations, and D is the surface dew point depression will be defined more specifically later. While these (°F). It has been suggested that wind velocity has an elements of the Fuel Moisture Component (FMC) exponential effect on the spread of fire among finer often act in cooperation, there are times when they fuels such as grass and chaparral, and that wind can are out of sync with one another due to the variability also have the same effect on fire spread as a fire in precipitation (frequency and amount) across burning upslope with little or no wind (Rothermel southern California in the winter. The relationship

1972). Dew point depression (T-Td) depicts the between these three components is expressed in the dryness at the surface well, and has a profound following equation: impact upon fuel conditions. Also, dew point depression can sometimes differentiate better [2] between warm and cold offshore events than relative 1 humidity can. In our dataset, it has been noted that larger dew point depression values (D >= 24°C) have mainly been associated with warm events. where DL is a Dryness Level index consisting of the While this may seem trivial, cold Santa Ana wind Energy Release Component (ERC) and the ten hour events (surface ambient temperatures < 16°C) are dead fuel moisture timelag (10-h). Live Fuel usually not associated with large fires. This may be Moisture (LFM) is a sampling of the moisture due in part to lower fuel temperatures because in content of the live fuels indigenous to the local those cases more time would be needed to reach the region, and G is the greenness of the annual grasses. ignition temperature. Another reason is that colder Currently we are making the assumption that all the events are sometimes preceded by precipitation either terms in equation [2] have equal weight, but further by a few days or by a few weeks which would cause study may lead to future modification. ERC is a fuels to be less receptive to new ignitions. These are relative index of the amount of heat released per unit the primary reasons why temperature was excluded area in the flaming zone of an initiating fire and is from equation [1] although it has been incorporated comprised of live and dead fuel moisture as well as temperature, humidity, and precipitation. While ERC indirectly through the use of D and in the fuels component that will be discussed in the following is a measure of potential energy, it also serves to section. Finally we note that while equation [1] bears capture the intermediate to long term dryness of the some resemblance to the FFWI, a comparison of fuels with unitless values generally ranging from 0- 100. The 10-hr dead fuel moisture timelag represents daily outputs of FFWI and LFP, revealed that LFP provides significantly greater contrast between Santa fuels in which the moisture content is exclusively Ana days and non-Santa Ana days. Therefore, these controlled by environmental conditions (Bradshaw et. al., 1983). Output values of 10-h are in g/g expressed results favored LFP as being the more appropriate equation for our purpose. as a percentage ranging from 0-35. In the case of the 10-h, this is the time required for the fuels (1/4” – 1” in diameter) to lose approximately two-thirds of their b) Large Fire Potential – Fuel Conditions initial moisture content (Bradshaw et al. 1983). Thus In addition to the meteorological conditions, large the DL index mainly serves to capture the dead fuel fire potential is also highly dependent on the state of moisture and has three unitless categories: 1 indicates the fuels. Given the complexity of the fuel fuels are moist, 2 represents average fuel dryness, environment (i.e. fuel type, continuity, loading, etc.), and a 3 indicates that fuels are drier than normal. we decided to focus more specifically on the moisture content of fuel conditions since that The observed LFM is the moisture content of live component plays a critical role in the spread of fuels, e.g. grasses, shrubs, and trees, expressed as a wildfires (Chuvieco et al. 2004). For our purpose, ratio of the weight of water in the fuel sample to the we have condensed fuel moisture into three oven dry weight of the fuel sample (Pollet and Brown

NDVI

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

Figure 5 ‐ Time series showing the seasonal variability of the Normalized Difference Vegetation Index (NDVI).

2007). Soil moisture as well as soil and air green-up usually occurs by December across temperature govern the physiological activity which southern California. During the green-up phase, results in changes in fuel moisture (Pollet and Brown grasses will begin to act as a heat sink, thereby 2007). LFM is a difficult parameter to evaluate preventing new ignitions and or significantly because of the irregularities associated with observed reducing the rate of spread among new fires. By late values. For instance, samples of different species of spring these grasses begin to cure with the curing native shrubs are normally taken twice a month by phase normally completed by mid-June. In equation various fire agencies across southern California. [2], G is a value that quantifies the said green-up and However, the sample times often differ between curing cycles of annual grasses. agencies and the equipment used to dry and weigh the samples may vary from place to place. In G is derived from the Moderate Resolution Imaging addition, sample site locations are irregular in Spectroradiometer (MODIS) NDVI data at a distribution and observations from these sites may be resolution of 250 meters for select pixels consisting taken sporadically. This presents a problem when we solely of grasslands. NDVI is further defined by red attempt to assess LFM over the region shown in and near-infrared (NIR) bands in the following figure 2. Apart from taking fuel samples, there are equation: several ways of estimating LFM using [3] meteorological variables, soil water reserve, solar radiation, etc. (Castro et al. 2003). One method uses satellite derived Normalized Difference Vegetation Index (NDVI) data and surface temperatures, as well as the Julian day, to form a linear regression equation where ρb= reflectance in band b (Clinton et. al., which approximates the LFM content of C. 2010). It can be shown that NDVI values for landanifer, a shrub species common in Spain and Southern California grasslands generally range from other parts of the Mediterranean (Chuvieco et al. about 0.25 (±0.05) to 0.75 (±0.05) for an average 2004). Since the climate and fuel conditions of southern California are similar to those of Spain, the authors will consider taking this same approach to estimate LFM rather than relying on the observed LFM values.

Following the onset of significant wetting rains, new grasses will begin to emerge in a process called green-up. While the timing and duration of this Table 1 ‐ Greenness (G) values associated with NDVI ranges. process fluctuates from year to year, some degree of

rainfall year (Fig. 5). There is evidence that NDVI is affected by soil color (Elmore et. al., 2000), which may explain the NDVI differences (±0.05) seen among the selected Southern California grassland locations.

G is given a rating of 0-5 based on NDVI data, where 0 is green and 5 is fully cured. When applying the methodology discussed by White (White et al. 1997) to the general range of Southern California grasslands, green-up is estimated to have occurred when NDVI exceeds 0.50. However, we have found that this value can be closer to 0.55 for some sites. Therefore, NDVI values greater than or equal to 0.55 Figure 6 ‐ Map showing maximum values of LFP (Eq. 1) are assigned a value of 0, or green. Furthermore, for every October from 1979‐2010. Solid contours NDVI values less than 0.35 are assigned a value of 5. indicate terrain height. This is because NDVI values are observed to be LFP value for each grid point in the model domain below 0.35 for all grassland sites during the dry for every October between 1979 and 2010. The result season when grasses are known to be fully cured. A was a composite map showing the areas where LFP linear relationship exists between NDVI for Southern is greatest (Fig. 6). The process was repeated for the California grasslands and fire occurrence. For this same time period substituting the month of January reason, the transition between green and fully cured for October (not shown). The results were similar (or vice versa) was given a rating of 1 to 4 in NDVI except the areas affected by LFP were larger and the increments of 0.05 (Table 1). values were higher overall. This is because offshore FMC modifies equation [1] in cases where fuels have wind events across southern California tend to not fully cured and are still inhibiting fire spread. become stronger in the winter due to more favorable Output values of FMC range from 0 to 1, where 0 atmospheric dynamics (Raphael 2003). The two represents wet fuels and 1 denotes dry fuels. This composite images were used to define what are called modifier can become so influential that it will greatly Santa Ana regions (or sub-regions) within each zone, reduce or even eliminate the potential for large fire which were further refashioned into rectangular occurrence despite favorable meteorological numbered boxes (Fig. 7). conditions for rapid fire growth. So the final equation for large fire potential becomes: Next, daily historical values of LFP were calculated for each zone, but rather than calculating LFP for [4] every grid point in the entire zone (which would include areas not affected by the wind event), LFP was calculated for grid points only within the boxed

4. Redefining the OFSI

The time and spatial problems with the OFSI that were previously mentioned have since been addressed through the use of the Weather Research and Forecasting (WRF) Advanced Research WRF (ARW) model, run by Robert Fovell at the Department of Atmospheric and Oceanic Sciences at UCLA. Using reanalysis data from the WRF-ARW at Figure 7 ‐ Map depicting Santa Ana regions (shaded) with 6 km resolution, we initially calculated a maximum numbered boxed areas. 6

areas of figure 7. The following equation was used to LFP for only one consecutive eight-hour time period calculate LFP at each grid point: may fail to capture the worst conditions of the day. The end result from equation [6] yielded a listing of

[5] daily LFP values per zone for months October-April from 1982-2011. ⋯ 8 Equation [2] for this same time period has yet to be calculated due to the absence of data. In order to 12 AM where LFP is an average complete the evaluation of FMC, ERC and 10-h must 1 AM 2 AM LFP value over an eight-hour be calculated at each grid point. At the time of this 3 AM 1 time period at grid point x. An 4 AM writing it is undetermined whether or not the 5 AM eight hour time period was traditional algorithms (Bradshaw et al. 1983) will be 6 AM chosen because that is ample 7 AM used or if a regression equation will be employed to 4 8 AM time for the finer fuels (i.e. 10- approximate these values. Also, algorithms for 9 AM 10 AM hr) to respond to the ambient calculating LFM and G have yet to be developed. 11 AM 2 atmospheric conditions. Once Final output from these variables is expected later 12 PM 1 PM an average LFP had been this spring or early summer, at which point an in 2 PM calculated for each grid point, 3 PM depth analysis will be performed. In this forthcoming 5 4 PM the following equation was analysis, daily LFP values for each zone will be used 5 PM 6 PM used to evaluate LFP for each to derive conditional probabilities of large fire 7 PM 3 zone: occurrence. Breakpoints within these conditional 8 PM 9 PM probabilities will ultimately lead to the final 10 PM [6] redefined index, which will likely be called the 11 PM “Santa Ana Wildfire Threat” (SAWT) index. Despite ⋯ the fact that FMC has yet to be calculated for the entire dataset, preliminary results from Zone 3 show how incorporating FMC into equation [1] modifies where LFP is an average of eight-hour averages at the LFP values which correlate well with observed each grid point in the sub-region boxes. It is fire activity. For example, consider the table below: important to note that equation [6] was calculated for five different eight consecutive hour time periods Note that LFP calculated for Zone 3 on 7 January (see table left) with the highest value chosen to 2003 had a value of 70, but when FMC was applied, represent each zone for the day. This is to ensure that the value decreased to 17. While there were ignitions the worst conditions are being captured on a daily on that particular date, no large fires occurred. The basis. For instance while most Santa Ana wind events case of 31 March 2005 is a more dramatic example of peak during the morning hours, some events can peak a situation in which the fuels were completely later in the day or at night depending on the arrival unsupportive of any fire activity, as the final LFP time of stronger dynamical support. Thus calculating value dropped to 0. Conversely, dates that had final

Date EQ1 ‐ LFPzone3 DL G LFM FMC EQ4 ‐ LFPzone3 Ignitions Large Fires Acres 1/7/2003 70 3 0 0.86 0.249 17 YN 1/1/2008 56 2 0 0.81 0.147 8 NN 10/22/2007 54 3 5 0.55 0.945 51 YY 9472 10/21/2007 51 3 5 0.55 0.945 48 YY197990 11/30/2006 48 3 5 0.58 0.920 44 YY 296 3/31/2005 48 1 0 1.19 0.000 0 NN 1/21/2010 46 1 0 0.70 0.043 2 NN 12/17/2004 45 2 0 0.76 0.164 7 YN 2/2/2005 43 1 0 0.90 0.011 0 NN 2/6/2006 41 2 4 0.68 0.595 24 NN 1/10/2009 41 2 0 0.66 0.203 8 YN 1/17/2008 40 2 0 0.87 0.130 5 YN

LFP values exceeding 40 did result in large fires when ignitions occurred.

Currently LFP is being computed operationally by UCLA using the 12 UTC WRF – ARW model run at

3 km to produce maps displaying maximum LFP values for the three zones (Fig. 8). Evaluation of this output is being performed by UCLA, Predictive Services, and SDG&E to compare forecast output with observed values. Expanded output is planned and proposed for the coming months which will include a comparison of forecasted values with normalized values and a breakdown of which component may be contributing the most toward the Figure 8 ‐ Map showing forecasted maximum LFP final output. (Eq 1) values.

months. The data from the survey and interviews 5. Social Science Impact will contribute to the upcoming second and third phases of the project. The survey results include data In order for the SAWT index to be a meaningful tool about WUI residents' most commonly used sources for the public, it is important that the social impact of for meteorological and emergency fire information such an index be studied in a detailed on a daily basis and during a wildfire event. This data manner. Understanding social behavior in this will drive the second phase of the project, which context will increase our awareness of the public’s focuses on interviews with media representatives ability to prepare for, respond to, and recover from from the identified media outlets. These interviews wildfires. This understanding is vital to the process of will be used to develop a better understanding of how meteorological product development, which if done meteorological and fire information is communicated correctly, will ensure that sustainable, comprehensive in the media and how it can be influenced to become communication is achieved. To this end, the Desert more consistent and relevant to WUI residents. Phase Research Institute (DRI) has been contracted to study three will utilize data from the risk perception scale the social aspect of this project and to make to help assess the most efficient number of levels in recommendations in helping condense and simplify the index, their associated definitions, the efficacy previously discussed output into a product that is of different modes and types of information, and relatable to the user. potential color schemes to use in the final presentation of the index. DRI is taking a multi-phased mixed method design with regard to the research methodology (Bergman Impacts from this research will govern the future of 2008; Morese and Neihause 2009). Phase one the SAWT index. In order to achieve maximum involved data collection in the form of a survey of success, a detailed communication campaign strategy Wildland Urban Interface (WUI) areas in southern will be deployed to educate stakeholders of the index. California (n=400) that assessed how meteorological It is our hope that this index will achieve the same and fire information is sourced, perceived, and level of success that the Saffir-Simpson scale has had processed by residents. The survey regarding the classification of hurricanes, in terms of instrument also included a scale to assess residents’ the ability to communicate a storm’s severity simply perceptions of wildfire risk (adapted from Trumbo et. and effectively to the public, the media, and to first al. 2012). In addition, in-depth interviews responders. on meteorological fire information use and needs were conducted with fire agencies (county, state, and federal) and will continue through the coming

6. Conclusion future research involving seasonal outlook predictions. As the WUI continues to expand across southern California, the source of ignitions will increase leading to a greater probability for large and destructive fires during Santa Ana wind events. This 7. Acknowledgements puts the public and firefighter safety at risk, thus the The authors would like to thank Jim Means for his increasing need to categorize such events in terms of helpful suggestions and input into this project. Many their effect on the fire environment. thanks also to Robert Fovell for his assistance with While the initial OFSI proved to be generally the WRF reanalysis data and his review of this paper. successful, its ability to capture the complexity of A special thanks to Yang Cao, and Scott Capps for all Santa Ana wind events was limited. This however led the hard work they put into producing the final to the development of a new methodology for a results. Finally, we appreciate the advice from Beth redefined index. Expressing large fire potential as a Hall, Tamara Wall, Mark Jackson, and Alex Tardy. function of wind velocity, dew point depression, and fuel moisture has allowed high resolution model and satellite derived variables to be incorporated into the 8. References index. This will specifically address the problems involving spatial coverage and fuel conditions innate Bergman, M.M. (ed.) 2008. Advances in mixed to the OFSI. methods research. Thousand Oaks, CA, Sage. Bradshaw, L.S., R.E. Burgan, J.D. Cohen, and J.E. Fuel moisture variables for the data period will be Deeming. 1983. The 1978 National Fire Danger obtained this summer, at which point a thorough Rating System: Technical Documentation. USDA analysis will be conducted to develop the final Forest Service; Intermountain Forest and Range product. A prototype product is scheduled to be Experiment Station, General Technical Report INT- released to a small test group in September. Full 169, Ogden, Utah. 44 pp. Castro, F.X., A. Tudela, and M.T. Sebastiá. 2003. product deployment is expected by September 2014. Modeling moisture content in shrubs to predict fire risk in Catalonia (Spain): Agricultural and Forest The use of social science will help in the Meteorology, 116, 49-59. understanding of how various survey recipients react Chuvieco, E., D. Cocero, D. Riaño, P. Martin, J. to meteorological and fire information. This Martínez-Vega, J. de la Riva, and F. Pérez. 2004. knowledge will eventually dictate what the final Combining NDVI and surface temperature for the product will become in terms of its content and estimation of live fuel moisture content in forest fire aesthetic. This last phase of the project is perhaps the danger rating: Remote Sensing of Environment, 92, 322-331. most important as the success of the index will be Clinton, N.E., C. Potter, B. Crabtree, .V Genovese, P. determined by what information is conveyed and how Gross, and P. Gong. 2010. Remote Sensing-Based it is presented. Time-Series Analysis of Cheatgrass (Bromus Tectorum L.) Phenology: Journal of Environmental The benefits of categorizing Santa Ana wind events Quality, 39, 955- 963. are multifold. Fire agencies and first responders, Elmore, A.J., J.F. Mustard, S.J. Manning, and D.B. private industry, the general public, and the media Lobell. 2000. Quantifying Vegetation Change in will have a clearer understanding of the severity of an Semiarid Environments: Precision and Accuracy of event based on the potential for large fires to occur. Spectral Mixture Analysis and the Normalized Difference Vegetation Index: Remote Sensing of Specifically, a more effective media response will Environment, 73, 87-102. result in the general population (particularly those Fosberg, M.A. 1978. Weather in wildland fire living within the WUI) being more proactive in its management: The fire weather index, paper response to an impending event. In addition, a presented at the Conference on Sierra Nevada climatology of Santa Ana wind events can be Meteorology, Am. Meteorol. Soc., South Lake developed based on this index which can be used in Tahoe, Calif.

Fovell, R.G. 2012. Downslope Windstorms of San Diego County; Sensitivity to Resolution and Model Physics: 13th WRF Users Workshop, June 2012. Hughes, M., and A. Hall. 2010. Local and synoptic mechanisms causing Southern California’s Santa Ana Winds: Clim, Dyn. 34:6, 847-857. Moritz, M.A., T.J. Moody, M.A. Krawchuck, M. Hughes, and A. Hall. 2010. Spatial variation in extreme winds predicts large wildfire location in chaparral ecosystems: Geophys Res Lett, 37, 1-5, doi:10.1029/2009GL041735. Morese, J.M. and L. Neihause. 2009. Mixed method design: Walnut Creek, CA, Left Coast Press Inc. Pollet, J. and A. Brown. 2007. Fuel Moisture Sampling Guide: Bureau of Land Management, Utah State Office, Salt Lake City, UT. Preisler, H.K., S-C. Chen, F. Fujioka, J.W. Benoit, and W.L. Westerling. 2008. Wildland fire probabilities estimated from weather model- deduced monthly mean fire danger indices: International Journal of Wildland fires, 17, 305- 316. Raphael, M.N. 2003. The Santa Ana Winds of California: Earth Interactions, 7, 1-13. Rolinski, T., D. Eichhorn, B.J. D'Agostino, S. Vanderburg, and J.D. Means. 2011. Santa Ana Forecasting and Classification: American Geophysical Union, Fall Meeting 2011, abstract #NH33B-1573. Rothermel, R.C. 1972. A Mathematical Model For Predicting Fire Spread In Wildland Fuels: Technical Documentation. USDA Forest Service; Intermountain Forest and Range Experiment Station, Research Paper INT-115, Ogden, Utah. 41 pp. Trumbo C.W., L. Peek, M. Mayer, H. Marlatt, B. McNoldy, and E. Gruntfest. 2012. A Multidimensional Examination of Hurricane Preparedness and Evacuation Intention: Poster presented to the Society for Risk Analysis, San Francisco, CA. White, M.A., P.E. Thornton, and S.W. Running. 1997. A Continental Phenology Model for Monitoring Vegetation Responses to Inter-Annual Climatic Variability. Global Biogeochem. Cycles, 11, 217- 234.